Jianfei Wang (ēå„é£) is a computer engineer and system architect working on autonomous driving systems. He envisions a future where moving atoms becomes as cheap and programmable as moving bits.
He previously designed Megviiās largeāscale training infrastructure, scaling it (prior to 2019) to more than 10,000 GPUs and 100 PB of data. In 2021 at Axera he designed an NPU that, as of 2025, remains among the most efficient1 edge accelerators for CNN and Transformer workloads.
He earned his B.S. in Computer Science and Technology from Tsinghua University in 2012.
Vision backbone benchmarks and LLM samples↩